Phylogeny, ancestral range reconstruction and partial taxonomic revision of Pacific Cyrtandra (Gesneriaceae)

Cyrtandra J.R. & G.Forster (Gesneriaceae) is one of the most widely dispersed plant genera in southeast Asia and the Pacific. Species concepts are variable and characters used to differentiate species are minimally useful between lineages. Molecular-based approaches for interpreting relationships between areas and for assigning taxonomic rank have been proposed. As part of this dissertation, I conducted three independent but related studies to address relationships in Cyrtandra: (1) I first used a preliminary phylogeny of Cyrtandra to compare four methods of ancestral range reconstruction: two developed for character-state reconstruction (Fitch parsimony and stochastic mapping), and two developed for ancestral range reconstruction (dispersal-vicariance analysis and dispersal-extinction-cladogenesis). The methods yielded conflicting results, dependent upon their respective assumptions. Likelihood-based methods allowed analytical interpretation of results useful in evaluating reconstructions. Dispersal-extinction-cladogenesis was the most applicable in my study, possibly due to its incorporating distance and timing of connections between areas. (2) I used molecular phylogenetic, diversification rates, and ancestral range analyses to construct a well-resolved evolutionary hypothesis for Cyrtandra. Results support a greater Fiji-Samoa region, corresponding with Takhtajan's 'Fijian Region,' as a major 'cross-roads' for Cyrtandra in the Pacific. I also compared existing taxonomy to my hypothesis to better understand its applicability. Current classifications are partially artificial; I suggest that sectional classifications should be revised to reflect monophyletic lineages. Also, species-level relationships should be closely studied in relation to these lineages. (3) Finally, I conducted a study/review of Hawaiian Cyrtandra as a model for phylogenetic-based revision in this genus. Morphological groupings, taxon distributions and taxonomy are well described for Hawaiian Cyrtandra but conflict with the current phylogenetic hypothesis. To stabilize taxonomy for future revision, a preliminary key to current sections is proposed and five species previously not treated are presented and classified. Results from this and the previous two studies indicate that a phylogenetics-based approach to taxon revisions is most appropriate for understanding evolutionary relationships in unwieldy genera such as Cyrtandra. Future studies in this genus will benefit from comprehensive, population-level molecular analysis in well-studied areas such as Hawai'i and from increased sampling in under-collected areas such as Fiji.